Study On The Treatment Of Air Pollution Data With SVM Regression Test Method

Air pollution has become a major issue in the current society.It not only endangers people’s health,but also brings irreversible damage to the natural environment of the world.Air pollution is mainly caused by human activities,and the major pollutants it generates are gradually increasing.Now the EPA is monitoring the data of pollutants,because we can use pollutant monitoring data to better formulate environmental protection programs.It can be seen that the pollutant data is very important.This is exactly what the paper put forward here,using the current data from Tianjin City’s air pollution data obtained from the Tianjin Environmental Protection Bureau as a basis to use the relationship between pollutants to horizontally forecast and supplement missing data.The test was conducted in the Matalab environment and the following main tasks were mainly completed:(1)Refer to the relevant literature,investigate and predict the methods of analysis and prediction of atmospheric pollution and data,and prove the advantages of the research content proposed in this paper;(2)Data pre-processing,and selecting SVM algorithm for data prediction training,research prediction accuracy;(3)PSO algorithm is used for parameter optimization,and it is expected to further improve its accuracy and provide an intuitive error graph for reference.The advantages of this study are that:1.Different from the usual longitudinal predictions,we propose to use the part of the data missing from lateral prediction of pollutants,ie,six pollutant data,and use five of them to predict the sixth pollutant value..2.The use of PSO to optimize the application of SVM in air pollution will greatly reduce the accuracy of prediction data while reducing human participation in calculations.Finally,experiments on real atmospheric pollution data demonstrate the effectiveness of the method.